Effects of White-Tailed Deer on Populations of an ......Effects of White-Tailed Deer on Populations...

995

Conservation Biology, Pages 995–1004Volume 12, No. 5, October 1998

Effects of White-Tailed Deer on Populations of an Understory Forb in Fragmented Deciduous Forests

DAVID J. AUGUSTINE*‡ AND LEE E. FRELICH†

*Department of Fisheries and Wildlife, University of Minnesota, 1980 Folwell Ave., St. Paul, MN 55108, U.S.A.†Department of Forest Resources, University of Minnesota, St. Paul, MN 55108, U.S.A.,email [email protected]

Abstract:

The effects of grazing by white-tailed deer (

Odocoileus virginianus

) on populations of

Trillium

spp.were examined in remnant, old-growth patches of the highly fragmented Big Woods forest ecosystem in south-eastern Minnesota. We conducted three separate studies involving an exclosure experiment, transplant exper-iments, and comparisons of

Trillium

populations among study sites. The highest grazing intensity was ob-

served where deer occurred at high overwinter concentrations (

z

25–35/km

2

); significantly lower grazingintensities occurred at low overwinter density (

z

5–10/km

2

). Deer focused their grazing on large, reproduc-tive plants; at sites with high deer density,

Trillium

population structure was skewed toward small plants, anddeer consistently caused over 50% reduction in reproduction during the growing season. Protection of indi-vidual plants from deer for two growing seasons resulted in dramatically increased flowering rates and sig-nificantly greater leaf area compared to control plants. No significant impact of current-year herbivory on re-production in the following year was detected. Nevertheless, flowering rates at one site with high overwinterdeer densities for at least the past 5 years suggest that the cumulative effects of grazing over several years canreduce reproduction in subsequent years. Transplant experiments with

Trillium grandiflorum

also showed thatdeer had significant effects on growth and reproduction where deer occur at high density. Our results suggestthat changes in landscape structure and local deer abundance have altered plant-deer relationships such thatgrazing can lead to the local extirpation of sensitive forbs such as

Trillium

spp. As a result, active, long-termmanagement of deer at low densities appears necessary for the conservation and restoration of fragmentedforest communities in eastern North America.

Efectos del Venado Cola Blanca en Poblaciones de

Trillium

spp. en Bosques Decidous Fragmentados

Resumen:

Los efectos de ramoneo por venado cola blanca (

Odocoileus virginianus

) en poblaciones de

Tril-lium

spp. fueron examinadas en parches remanentes de bosque maduro del ecosistema altamente fragmen-tado del bosque Big Woods en el sureste de Minnesota. Realizamos tres estudios separados que involucraronun experimento en un encierro, experimentos de transplante y comparaciones de poblaciones de

Trillium

en-tre sitios de estudio. La intensidad del ramoneo más alta se observó donde los venados se concentraron parapasar el invierno (

z

25–35/km

2

), mientras que intesidades menores de ramoneo ocurrieron a bajas den-sidades durante el invierno (

z

5–10/km

2

). Los venados enfocaron su consumo en plantas reproductivasgrandes; en sitios con alta densidad de venados la estructura poblacional de

Trillium

estuvo sesgada a plantaspequeñas y los venados consistentemente ocasionaron una reducción mayor al 50% en la reproducción du-rante la temporada de crecimiento. La protección de plantas individuales de la depredación de los venadospor dos temporadas de crecimiento resultó en un dramático incremento en la tasa de florecimiento y unacobertura de hojas significativamente mayor comparando con plantas control. No se detectó un impacto sig-nificativo de la herbivoría del año en curso en la reproducción del año siguiente. Sin embargo, las tasas deflorecimiento en un sitio con alta densidad de venados durante el invierno por al menos cinco años consecu-tivos sugiere que el efecto acumulativo del ramoneo a lo largo de varios años puede reducir la reproducción

‡

Current address: Biological Research Laboratories, Syracuse University, 130 College Place, Syracuse, NY 13244–1220, U.S.A., email [email protected] submitted June 4, 1997; revised manuscript accepted November 18, 1997.

996

Deer Impacts on

Trillium

spp. Augustine & Frelich

Conservation BiologyVolume 12, No. 5, October 1998

en años subsecuentes. Experimentos de transplantes con


mostraron también que los ve-nados tienen un efecto significativo en el crecimiento y reproducción en sitios donde los venados ocurren elaltas densidades. Nuestros resultados sugieren que los cambios en la estructura del paisaje y la abundancialocal de venados han alterado las relaciones planta-venados a un punto en el que el ramoneo podría con-ducir a la extirpación local de especies sensitivas como lo son

Trillium

spp. Como resultado, aparenta ser nec-esario el manejo activo a largo plazo de venados a bajas densidades para la conservación y restauración de

comunidades de bosques fragmentados en el este de Norteamérica.

Introduction

Human-induced changes in the abundance of mamma-lian herbivores, through changes in hunting practices,predation pressure, and exotic introductions, can criti-cally alter the structure and dynamics of ecosystemsworldwide (Howard 1966; Harper 1969; Spatz & Muel-ler-Dombois 1973; Leuthold 1996; McShea & Rappole1997). One documented consequence of increased white-tailed deer densities in eastern North America is the del-eterious effect on populations of sensitive tree speciesand the resulting potential for local species extirpations(Anderson & Loucks 1979; Frelich & Lorimer 1985; Alli-son 1990). Although extensive research has examinedwinter browsing effects on forest communities (Graham1952; Beals et al. 1960; Tilghman 1989; Strole & Ander-son 1992; Anderson & Katz 1993), considerably less at-tention has been paid to growing-season deer herbivoryand its potential effects on understory herbaceous plantcommunities (Miller et al. 1992; Anderson 1994; Bal-gooyen & Waller 1995; Rooney 1997). As managers con-tinue to face difficult decisions concerning the manage-ment of deer populations, often with limited availableinformation, a better understanding is needed of the ef-fects deer exert on native plant communities.

Alverson et al. (1988, 1994) suggested that human-in-duced changes in the distribution of plant communitiescan alter plant-herbivore relationships and can poten-tially cause significant herbivore impacts within the newlandscape. In Minnesota, dramatic changes in the struc-ture of forest landscapes have changed the distributionand abundance of white-tailed deer (Berner & Simon1993). In southeastern Minnesota, some 8000 km

2

ofcontiguous mesic deciduous forest in the presettlementlandscape dominated by elms (

Ulmus americana

and

Ulmus rubra

), sugar maple (

Acer saccharum

), andAmerican basswood (

Tilia americana

) (Daubenmire1936; Grimm 1984) have been almost entirely convertedto agricultural and suburban land uses. Today, the onlyremaining fragments of this forest type, commonlytermed Big Woods forest, are on the order of 8–32 ha(Jakes 1980; Vasilevsky & Hackett 1980), and conserva-tion plans focus on protecting existing old-growth frag-ments, restoring forest corridors between fragments,

and restoring characteristic species. Deer were extir-pated or extremely rare in this region of the state fromthe late 1800s through the 1920s, and counties in south-eastern Minnesota were still closed to hunting in mostyears during the 1950s and 1960s to promote popula-tion recovery (Berner & Simon 1993). Over the pastthree decades, deer populations in southeastern Minne-sota have increased steadily (Berner & Simon 1993; Dex-ter 1996).

Recent surveys conducted in four old-growth BigWoods remnants in southeastern Minnesota documentedselective deer foraging patterns on understory forbs (Au-gustine 1997). Species in the genus

Trillium

were con-sistently grazed to a greater degree than most forb spe-cies at different levels of

Trillium

abundance and deerdensity. Although studies have suggested that deer caneliminate selected herbaceous species from forestpatches (e.g., Alverson et al. 1988; Anderson 1994), thishas not been experimentally addressed. Our objectiveswere therefore to examine (1) the effect of deer her-bivory on the size structure of

Trillium

populations atvarious local deer densities, (2) the effect of deer her-bivory on the reproductive potential of

Trillium

popula-tions, and (3) the potential influence of deer on restora-tion of

Trillium

to the forest community.

Methods

Study Area

Studies were conducted at three old-growth Big Woodsremnants located in Rice County (44

8

15

9

N, 93

8

20

9

W)and Hennepin County (45

8

N, 93

8

30

9

W) in southeasternMinnesota. The sites are dominated by sugar maple,American basswood, and elms (

.

80% of relative domi-nance by basal area); they exhibit an all-aged distributionof tree sizes and contain large (50–100 cm diameter atbreast height [dbh]) individuals of the three dominanttree species (Augustine 1997). Ironwood (

Ostyra vigi-nana

) is an important subcanopy species. Sites containloamy soils developed from glacial moraines or silty soilsdeveloped from loess-covered glacial till (Grimm 1984).The three forests were selected to include two sites with


Augustine & Frelich Deer Impacts on

Trillium

spp.

997

relatively high local deer densities and one site with acomparatively low local deer density based on 1993–1994 aerial counts.

Species Descriptions

Trillium cernuum

and

Trillium flexipes

are long-lived,nonclonal forbs found in the understory of mesic decid-uous forests in eastern North America (Gleason & Cron-quist 1991). In Minnesota, specimens have been col-lected throughout the Big Woods region (Ownbey &Morley 1991). In many remaining Big Woods forests inMinnesota, including all three study sites,

T. cernuumand T. flexipes

occur sympatrically, and individuals withhybrid characteristics are commonly observed (Rogers1981; D.J.A., personal observations). As a result, we con-sidered these species as a single

T. flexipes–T. cernuum

complex.


is a long-lived, nonclonal forbdistributed throughout rich, broad-leaved deciduous for-ests in eastern North America (Gleason & Cronquist1991).


has been collected fromdeciduous forests throughout central Minnesota andfrom stands south, east, and west of the Rice Countysites included in this study (Ownbey & Morley 1991). A1972 survey of a mature Big Woods forest fragment ap-proximately 5 km north of one of the Rice County sitesincluded in this study also documented the presence of

T. grandiflorum

(N. Falkum, personal communication).


does not occur naturally in anyof the study sites at this time.

Individual

Trillium

plants consist of a single stem(rarely two) supporting a whorl of three leaves in thelarger size classes and a single leaf in juvenile plants.


seeds germinate in the springand produce adventitious roots (no leaves) in the firstgrowing season, a cotyledon leaf in the second growingseason, and a single leaf in the third growing season(Hanzawa & Kalisz 1993). In Minnesota all three

Tril-lium

species reach anthesis in late May through earlyJune, and flowering plants begin to develop fruits inJune. Both

T. grandiflorum

and

T. flexipes

exhibit a sim-ilar population size structure with comparatively highmortality rates of seeds and juvenile size classes and lowmortality for larger size classes (Kawano et al. 1986;Ohara & Utech 1988).

Study Design

Three separate studies were used to examine the effectsof deer on

Trillium

populations. First, we examineddeer grazing intensity and the effect of grazing on popu-lation size structure at the two high-deer-density sites(H1 and H2) and the one low-deer-density site (L1). Sec-ond, we used small deer exclosures placed around indi-vidual

Trillium

plants, each paired with an unexclosed

control plant, to measure the magnitude of the effects ofdeer on growth and reproduction at a site with highdeer density. Finally, we examined potential deer influ-ences on plant community restoration efforts by estab-lishing transplanted populations of

Trillium grandiflo-rum

at one site with high deer density and one with lowdeer density. This species was chosen because

T. gran-diflorum

is palatable to deer (Anderson 1994; Bal-gooyen & Waller 1995), is absent from many extant BigWoods forests, and is known to have occurred through-out the Big Woods region in Minnesota. The two siteswhere

T. grandiflorum

was transplanted are located to-ward the southern edge of the species’ range and hencemay represent stressful growing conditions where herbi-vore effects could be especially severe. Randomly cho-sen

Trillium

transplants were individually protectedfrom deer at each site, creating a 2

3

2 factorial experi-ment with high and low grazing pressure and protectedand unprotected plants.

Deer Density Estimation

Three methods were used to estimate seasonal deer den-sities at each study site. First, winter density was mea-sured by deer pellet-group counts. Pellet groups werecounted within a 4-m radius of 46–50 fixed samplingpoints at each site immediately following snowmelt (14–20 March in 1995 and 1–12 April in 1996). Because leaf-fall in October creates a relatively uniform layer of litter,late-winter pellet-group counts represent accumulationonly for the late fall and winter months, providing an in-dex of overwinter deer presence within the stand.Counts were converted to relative estimates of deer persquare kilometer based on a 150-day deposition periodassuming 13 pellet groups produced per deer per day(Eberhardt & van Etten 1956). Density differences be-tween sites were compared by means of likelihood ratiotests for log-linear models (Agresti 1996).

We used aerial counts to obtain a second estimate ofwinter deer density at each study site as a check on pel-let count estimates (White 1992; Jordan et al. 1993).Aerial counts were conducted in January 1996, and thearea flown at each site included the mature Big Woodsstand and a surrounding mosaic of second-growth for-est, wetlands, and shrubland. Winter deer density fromaerial counts was expressed as the number of deercounted, uncorrected for observer bias, divided by thetotal area of permanent winter cover, defined as forests,wooded wetlands, and shrubland, within the area flown(Augustine 1997). An analysis of aerial count versus pel-let count estimates at six Big Woods forests in southeast-ern Minnesota showed that both methods provide con-sistent results for deciduous forests, except in severewinters where alternative winter habitat is locally avail-able (Augustine 1997).

998

Deer Impacts on

Trillium

spp. Augustine & Frelich


Deer density during spring and summer was measuredat each site with three automated cameras attached toinfrared deer monitors (Non-typical Engineering, GreenBay, Wisconsin, U.S.A.), which were moved to new, ran-domly located positions every 7 days (Augustine 1997).Because we had only six monitors, L1 and H2 were sam-pled in 1995, and H1 was sampled in 1996. Data ob-tained from camera monitors were analyzed in terms ofthe number of deer photographed per week.

Grazing Intensity and

Trillium

Size Structure

We sampled naturally occurring

Trillium cernuum

and

T. flexipes

at L1, H1, and H2 using fixed, uniformlyspaced 50-m transects. A 2-m transect width was sam-pled at L1 where

Trillium

was abundant, and a 4-mwidth was used at H1 and H2. Transects were first sam-pled during 5–9 May 1995, when plants were emergingfrom the ground and each

Trillium

stem was markedwith a numbered aluminum tag. At this time, the repro-ductive status (flowering or nonflowering) of each plantwas recorded. Because

Trillium

was extremely rare atH2, few plants were observed along transects. We there-fore conducted extensive understory searches at thissite and marked all plants found between 14 April and15 May 1995.

Marked plants were re-surveyed during 19–22 June(after

Trillium

anthesis) to record reproductive status,whether each plant was grazed by deer, any other herbi-vore damage, stem height, and the length and width ofone randomly selected leaf for plants that were not com-pletely defoliated. Deer grazing always resulted in 100%defoliation and was recognized by the rough cut of thestem, typically at a height of 15–30 cm. Plants were re-checked for deer grazing in August 1995.

In 1996 marked plants at each site were surveyed dur-ing 5–8 May and 17–20 June to record the same mea-surements as in 1995. In addition, because relatively lit-tle deer grazing occurs before early May, the length ofthe emerging leaf bundle of each plant was recorded inthe May survey to obtain a pre-grazing distribution ofplant sizes for each population. Plants were recheckedfor deer grazing in late June and late July 1996.

Trillium

individuals sampled at each site may not beindependent with respect to growth if they are locatedclose to one another. For data analyses any time twomarked plants were located within 20 cm of one an-other, we deleted one randomly selected plant in thepair from the sample; all reported sample sizes and anal-yses are based on this subset of plants.

For

Trillium

species, Kawano et al. (1986) found thatleaf area is a good measure of biomass and hence thegrowth stage of an individual. We estimated leaf area inthe field as plant leaf area

5

leaf length

3

leaf width

3

0.5

3

number of leaves. We measured true leaf area fora sample of 30

T.

grandiflorum

and 30

T.

cernuum

–

T.

flexipes

using an Agvision Monochrome System for LeafAnalysis to establish whether this method provides a re-liable estimate of individual leaf area.

Trillium

Exclosure Experiment and

T. grandiflorum

Transplant Experiments

To examine the effects of deer on individual growth andreproduction in a naturally occurring

Trillium

popula-tion, site H1 was searched during 9–10 May 1995 to findand mark 50

Trillium

with leaf length of at least 4 cm.Plants were paired (25 pairs) based on size, reproduc-tive status, and location, and one plant in each pair wasprotected with an individual welded-wire deer exclo-sure. Plants were monitored for reproduction, deer graz-ing, other herbivore damage, stem height, and thelength and width of one randomly selected leaf on 9–10May 1995, 27 June 1995, 7 May 1996, and 19 June 1996;they were checked for deer grazing on 17 August 1995and 26 July 1996.

We transplanted 120

T.

grandiflorum

rhizomes to siteL1 and 120 rhizomes to site H2 in August 1994. Individu-als were planted along transects at a minimum spacingof 3 m to ensure that each represented an independentobservation. At both sites, 40 transplants were randomlyselected and each plant was protected with a separate,welded-wire deer exclosure. Plants were monitored forreproductive status, deer grazing, other herbivore dam-age, stem height, and the length and width of one ran-domly selected leaf on 11–12 May 1995, 17–18 June1995, 11–12 May 1996, and 17–18 June 1996; they wererechecked for deer grazing on 26–27 August 1995 and25–26 July 1996.

Results

Deer Densities

Winter density based on pellet-group counts was signifi-cantly lower at L1 than at the high deer sites in both sea-sons of both years (Fig. 1; likelihood ratio tests,

D

G

2

.

12.25, df

5

1,

p

,

0.001, for all comparisons). Pellet-group counts in 1995 showed winter concentrations of24–31 deer/km

2

at H1 and H2 compared to 4 deer/km

2

at L1 (Fig. 1). Aerial counts showed an overwinter con-centration of 25–36 deer/km

2 of permanent cover at H1and H2 and 11 deer/km2 at L1 (Fig. 1). The pellet countestimate at H1 was significantly lower in 1996 than 1995(Fig. 1), but aerial counts conducted in both years at thissite showed a constant density (23.4 and 25.3 deer/km2). The change in pellet density was likely due to in-creased use of two conifer patches and a large south-fac-ing slope adjacent to site H1 during the more severewinter conditions in 1996. Overall, combined estimatesof winter deer density show higher local overwinter


Augustine & Frelich Deer Impacts on Trillium spp. 999

concentrations at H1 and H2 than at L1 during the studyperiod.

Winter aerial counts have been conducted intermit-tently over the past 8 years at H2 and L1 and every win-ter for the past 5 years at H1 (J. Moriarty and J. Vorland,personal communication). High deer densities havebeen present at H1 for at least the past 5 years, whereasthe current density at H2 is the result of a rapid increasebetween 1989 and 1993. Density at L1 has been lowover the past 8 years.

The automated cameras provided a sensitive index ofgrowing-season deer density, which was approximatelythree times higher at H2 than at L1 in 1995 (DG2 5 14.6,df 5 1, p , 0.001) and four times higher in 1996 at H1than the 1995 index for L1 (DG2 5 31.3, df 5 1, p ,0.0001; Fig. 1).

Grazing Intensity and Trillium Size Structure

We examined grazing intensity and population sizestructure for a sample of 143, 28, and 76 marked plantsin 1995 and 164, 47, and 90 marked plants in 1996 at L1,H1, and H2, respectively. Most grazing on Trillium oc-curred between early May and the second sampling pe-riod in late June, except at H2 in 1995 (Table 1). Grazingintensity was significantly different among sites in both

years (Table 1; x2 5 32.15 and 27.04, df 5 2, p ,0.0001). The highest grazing intensity occurred at H2 inboth years and was two to five times higher than overallgrazing intensities at both L1 and H1. No differences inoverall grazing intensity were detected between L1 andH1 in 1995 or 1996 (Table 1; x2 5 0.29, p 5 0.59; x2 50.6, p 5 0.80). In addition to deer grazing, some plantswere defoliated by lepidopteran larvae that sever thebase of the stem. At L1, H1, and H2 respectively, lepi-dopterans damaged 0.0, 0.0, and 6.6% of plants in 1995and 0.0, 0.6, and 4.4% of plants in 1996.

We examined two measures of Trillium populationstructure: the distribution of plant sizes within a popula-tion and the proportion of flowering plants in a popula-tion. Leaf length and width measurements provided ahighly significant predictor of true leaf area (r2 5 0.99,F 5 8512, p , 0.0001). In addition, the size class of theemerging leaf bundle (in 0.5 cm increments) measuredin early May was a significant predictor of the leaf areaof ungrazed plants in June after anthesis (weighted re-gression, r2 5 0.96, F 5 239, p , 0.0001). Measure-ments of leaf bundle length in early May were used toconstruct a pre-grazing plant-size distribution (#1%were grazed before May sampling), and leaf area esti-mates from late June were used to construct a post-graz-ing size distribution for each of the three populations.

No significant differences were detected among thesize-class distributions of the three populations in earlyMay (one-tailed K-S statistics #0.06, p $ 0.8 for all com-parisons; Fig. 2). By late June (postanthesis and postgraz-ing), significant differences in the size distributions ofthe three populations were observed (Fig. 2). At H1 thepopulation consisted predominantly of individuals withleaf area under 80 cm2, which differed significantly fromL1 (K-S statistic 5 0.28, p 5 0.05). The distribution at H2was shifted toward an increased frequency of smallerplants in the population but was not statistically differ-ent from that of the other sites (K-S statistics ,0.21, p .0.26). Although no early May measurements were madein 1995, the distribution of plant sizes in late June 1995at each site was similar to that observed in June 1996.

In early May, prior to the majority of deer grazing, theproportion of the plants flowering in each populationprovides another measure of the population’s structureprior to the within-season effects of grazing. In earlyMay of 1995 and 1996, the flowering rate at H1 (4% and15%) was significantly lower than that at both L1 (27%and 34%) and H2 (32% and 38%) (x2 $ 7.0, p # 0.008 forall comparisons). No difference was detected in the Mayflowering rate between H2 and L1 in either year (x2 #0.6, p $ 0.40 ).

Flowering individuals experienced a higher rate ofgrazing than the overall Trillium population (floweringand nonflowering plants) at all three study sites (Table1). Grazing intensity on reproductive plants was signifi-cantly greater at H2 than L1 in both years (Table 1; x2 5

Figure 1. Seasonal deer densities at three study sites, estimated with three methodologies. Winter density was measured by (1) pellet-group counts converted to an estimate of deer/km2 61 SE, assuming 13 pellet-groups per deer per day and (2) aerial counts. Rela-tive deer density estimates for spring-summer are ex-pressed as the number of deer photographed per week 61 SE and were measured at H1 and L1 in 1995 and H2 in 1996.

1000 Deer Impacts on Trillium spp. Augustine & Frelich


7.57 and 7.14, p , 0.007). The sample of reproductiveplants was extremely small at H1 due to their rarity; 4 of 8marked reproductive plants were grazed in 1995 and 10of 13 plants were grazed in 1996. The 1996 grazing ratewas significantly higher than the proportion of reproduc-tive plants grazed at L1 (Table 1; x2 5 6.42, p 5 0.01).

To examine the effects of grazing in 1995 on Trilliumreproduction in 1996, we compared plants that flow-ered and were grazed by deer in 1995 to plants thatflowered and were ungrazed in 1995 for data pooledacross all study sites. In early May 1996, flowering rateswere not significantly different between the grazed andungrazed sample (80%, n 5 25 versus 90%, n 5 44 re-spectively, x2 5 1.67, p 5 0.20). Given the sample sizes,the chance of detecting a 0.35 difference in floweringrates at alpha 5 0.1 was 90%.

Effects of pollinator limitation on these populationsappeared to be minimal because 87–100% of ungrazedflowering plants were developing fruits by the late Junesampling dates at all sites.

Response of a Trillium Population to Protectionfrom Herbivory

The exclosure experiment at H1 revealed a dramatic re-sponse of Trillium to release from herbivory after onlytwo growing seasons. At the beginning of the experimentin May 1995, plant size was similar for exclosed ( 543.86 cm2) versus unexclosed ( 5 45.49 cm2) plants(paired t test, t 5 0.54, p 5 0.3), and 28% of the plants inboth experimental groups were flowering. In 1996, after1 year of protection from deer herbivory, the leaf area ofunprotected plants in early May was significantly lowerthan the leaf area of paired, protected plants (26.8 cm2

versus 37.1 cm2, t 5 2.72, p 5 0.007). After two grow-ing seasons, the flowering rates of protected versus un-protected plants diverged dramatically, such that byJune 1996 the flowering rate of protected plants was 19times greater than the rate for unprotected plants (Fig.3; x2 5 27.0, p , 0.0001).

xx

Within a growing season, plants can be defoliated bydeer, mechanical damage to the stem, and lepidopteranlarvae. During both growing seasons, the proportion ofplants that were not defoliated by the late June samplewas significantly higher for protected than for unpro-tected plants (Fig. 3; 1995: x2 5 7.71, p 5 0.006; 1996:x2 5 22, p , 0.0001). These differences reflect the factthat deer grazed 36% and 44% of the unprotected sam-ple in 1995 and 1996, respectively. For protected andunprotected plants combined, 12% and 8% experiencedlepidopteran damage in 1995 and 1996, respectively,with no significant difference between treatments (p $0.8).

Trillium grandiflorum Transplant Experiments

In early May 1995, 58 unprotected and 19 protectedtransplants emerged at L1, and 56 unprotected and 18protected transplants emerged at H2 following highoverwinter mortality at both sites. At this time meanplant leaf area was similar inside and outside exclosuresat both sites (t # 1.20, p $ 0.24). In 1995, deer grazingintensity on unprotected plants was significantly greaterat H2 than at L1 (Table 2; x2 5 14.9, p , 0.0001). Nograzing on transplants was observed after late June. Theoverall proportion of transplants damaged by lepidopteranlarvae was also significantly greater at H2 (31.1% versus7.9%, x2 5 12.92, df 5 1, p 5 0.0003). At H2, however,the proportion of transplants retaining leaves throughanthesis was significantly greater for protected plants(available to lepidopteran larvae only) than for unpro-tected plants (available to deer and lepidopteran larvae)(x2 5 7.4, df 5 1, p 5 0.007). No significant differencewas detected between protected and unprotected plantsat L1 (x2 5 0.09, df 5 1, p 5 0.76).

Following 1 year of protection from deer herbivory,mean leaf area in early May 1996 at H1 was significantlygreater for protected than unprotected plants (t2 5 2.49,p 5 0.027). Grazing intensity decreased from 1995 to1996 at H2 and increased at L1, such that in 1996 there

Table 1. Trillium density and deer grazing intensity at three study sites in southeastern Minnesota.

Grazed (%)

Site*Trillium stems/

10 m2 6 SEOverall

growing season Emergence-anthesis Post-anthesisReproductive

plants grazed (%)

1995Low deer density (L1) 1.5 6 0.4 10.5 10.5 0.0 23.7High deer density 1 (H1) 0.07 6 0.05 7.1 7.1 0.0 —High deer density 2 (H2) 0.02 6 0.01 42.1 18.4 23.7 58.3

1996Low deer density (L1) 2.0 6 0.5 16.5 14.6 1.8 35.7High deer density 1 (H1) 0.07 6 0.05 19.1 12.8 6.4 76.9High deer density 2 (H2) 0.01 6 0.06 45.6 40.0 5.6 64.7

*Sites H1 and H2 had 3–4 times higher winter and growing-season densities than L1, based on aerial counts, pellet counts, and infrared cameramonitors (Fig. 1).



was no significant difference in the proportion of unpro-tected Trillium grazed (x2 5 0.99, p 5 1.0). In 1996 atH1, 35% of the unprotected plants disappeared due tounknown causes, whereas only 15% disappeared at L1,possibly resulting in a greater underestimate of deergrazing intensity at high deer density. All observed deergrazing occurred before late June. In 1996 the propor-

tion of plants damaged by lepidopteran larvae was lowat both sites (1.8% and 5.6%, respectively). Survival ratespast anthesis in 1996 were higher inside exclosures thanoutside at both low (Table 2; x2 5 4.94, p 5 0.03) andhigh deer density (x2 5 2.86, p 5 0.09).

Flowering rates of transplants were similar inside andoutside exclosures at the beginning of the experiment inearly May 1995 at both low deer density (94% versus84%, x2 5 1.34, p 5 0.25) and high deer density (78%versus 80%, x2 5 0.04, p 5 0.84). No significant differ-ence was detected between the proportion of trans-plants flowering in the post-anthesis sample (late June)for protected versus unprotected plants at L1 in 1995(Table 2; x2 5 0.00, df 5 1, p 5 0.97) or in 1996 (52.9%versus 38.2%, x2 5 1.16, p 5 0.28). At H2 a significantlygreater proportion of transplants was flowering at anthe-sis inside than outside exclosures in 1995 (x2 5 4.32,df 5 1, p 5 0.04) and 1996 (x2 5 5.1, df 5 1, p 5 0.02).

Discussion

Deer Impacts on Natural Trillium Populations

The effects of herbivory on a plant population dependon the degree to which plants are defoliated and the ef-fect of defoliation on individual reproduction and growth.Deer effects on individual Trillium are especially severebecause one bite causes 100% defoliation and loss of theflower or fruit from reproductive plants, no regrowthoccurs within that growing season, and Trillium do notreproduce clonally. At two forests where hunting hasbeen prohibited for more than 5 years and deer densitiesare high, deer consistently removed more than 50% ofreproductive individuals from the population each year,caused reductions in plant size, and altered the distribu-tion of plant size-classes in the population. In contrast,at a site where deer are hunted on an annual basis, deerdensity was three to four times lower, and grazing had

Figure 2. Cumulative frequency distributions of plant sizes in three Trillium populations experiencing differ-ent levels of deer herbivory. The early May distribution is based on the length of emerging leaf bundles (cm), whereas the late-June distribution is based on esti-mated leaf area (cm2) of ungrazed plants.

Figure 3. Trends in the flowering rate and survival within a growing season of naturally occurring Tril-lium at site H2 for plants protected from deer her-bivory versus those of unprotected plants.



significantly less severe effects on Trillium reproduc-tion, growth, and population structure. Although manyfactors may have led to the current low Trillium densi-ties at the sites with high deer density, the documentedlevels of grazing on flowering plants indicate that highlocal deer densities are preventing Trillium populationsfrom recovering. The exclosure experiment at H1 sup-ported this conclusion and demonstrated that changesin plant size and reproductive rates for unprotectedplants can be dramatically reversed when deer her-bivory is eliminated for only 2 years (Fig. 3). Observedseed-set rates ($87%) are higher than reported for T.flexipes populations elsewhere (Ohara & Utech 1988),indicating that pollination is not limiting reproduction.

In addition to the within-growing season reduction inTrillium reproduction, herbivory can further affect pop-ulations if defoliation in one growing season reduces re-production in subsequent years. The lack of a significantreduction in 1996 flowering rates for grazed comparedto ungrazed plants that were reproductive in 1995 indi-cates that one herbivory event does not have long-termeffects on Trillium reproduction. Partial defoliation(50%) of T. grandiflorum without removal of the flowerresults in the allocation of resources to sexual reproduc-tion at the expense of future growth and reproduction(Lubbers & Lechowicz 1989). The lack of reduced fu-ture reproduction observed in our study may result fromthe leaf biomass lost to deer being offset by the conser-vation of energy normally expended on seed productionbecause deer also remove the flower. This result sug-gests that Trillium individuals can be highly resilient todeer grazing over the short term (1–2 years) such that,following a reduction in deer densities, Trillium repro-ductive rates can rapidly return to levels observed at lowdeer density.

Nevertheless, the cumulative effects of herbivory overseveral growing seasons could cause long-term reduc-tions in reproductive success. Comparison of the Tril-lium populations at the two high-deer-density sitesshowed that, although both populations experiencehigh grazing intensity on large, reproductive plants, H1contains a significantly lower proportion of floweringplants than H2 in early May before most deer grazingtakes place. This likely reflects differences in deer densi-

ties over the past 5–8 years. At H2, high densities havebeen present only for the past 2–3 years, following arapid increase in the local deer population between1989 and 1993. At H1 aerial counts show a consistentlyhigh and slightly increasing local deer density from 1992to 1996. Lower pregrazing flowering rates at this sitesuggest that cumulative, long-term grazing impacts onreproduction occur when high local deer densities aremaintained for at least 5 years.

Although no baseline demographic data are availablefor T. cernuum or T. flexipes, the population structureof T. flexipes as determined by size-class analysis (Ohara& Utech 1988) is similar to that of T. grandiflorum(Kawano et al. 1986). For a T. grandiflorum populationin Michigan, Hanzawa and Kalisz (1993) found that indi-viduals ranged from 1 to 30 years old and the minimumage of reproductive plants was 17 years. This suggeststhat, if grazing intensities observed at sites with highdeer density continue for more than 15–20 years, deercan drive a population to local extirpation. Conversely,the relatively long generation time of Trillium will se-verely limit the rate at which population densities canrecover following a reduction in deer density.

In addition to direct grazing effects on Trillium, com-munity-level grazing patterns may affect Trillium estab-lishment, growth, and survival. Previous work showedthat Trillium spp. are grazed to a greater degree thanmany common forb species (Augustine 1997). At bothsites with high deer density, Trillium was extremelyrare in the understory (Table 1), but it still experiencedhigh grazing intensity. Forb species that are less palat-able and more tolerant of grazing occur at comparativelyhigh densities in the forest understory and likely com-pete for light and establishment sites. Equally importantwas the observation that high grazing intensities aremaintained on reproductive Trillium individuals evenwhere they are extremely rare in the overall plant com-munity. Because the preference of deer for Trillium hasalso been documented in Illinois (Anderson 1994) andnorthern Wisconsin (Balgooyen & Waller 1995), effectsobserved in this study may occur across a broad geo-graphic region.

This community-level grazing pattern can result instrong negative deer impacts on sensitive species while

Table 2. Grazing, growth, survival, and reproduction of T. grandiflorum transplants at two study sites in southeastern Minnesota.*

Low deer density site (L1) High deer density site (H2)

unprotected protected unprotected protected

T. grandiflorum transplant characteristics 1995 1996 1995 1996 1995 1996 1995 1996

Grazed (%) 3.4 12.8 — — 30.4 14.7 — —Surviving past anthesis (%) 86.2 61.7 83.3 89.5 17.9 41.2 50.0 65.0Reproductive post-anthesis (%) 63.9 38.2 64.7 52.9 8.9 14.3 33.3 38.9Early May leaf area (cm2) 37.8 43.4 42.4 41.0 43.5 38.6 50.8 65.4

*Site H2 had 3–4 times higher winter and growing-season densities than L1, based on aerial counts, pellet counts, and infrared camera moni-tors (Fig. 1).



the overall groundlayer remains green. Negative impactsmay extend to many long-lived understory forbs that,like Trillium, can have most leaf area and reproductivestructures removed in a bite, that possess no capacityfor regrowth after grazing, and that are highly palatableto deer (e.g., Clintonia, Maianthemum, Polygonatum,Sanguinaria, Smilacina, Smilax, and Uvularia; Bal-gooyen & Waller 1995; Augustine 1997; Rooney 1997).This contrasts with the common situation in which deerimpacts are recognized and responded to by managerswhen a “browse line,” characterized by the complete ab-sence of vegetation within the groundlayer, begins todevelop.

Deer Impacts on Plant Community Restoration

Conservation plans for Big Woods remnants call for therestoration of native species lost during forest fragmen-tation. In remnants protected from direct human influ-ences, lack of deer management can still impede restora-tion efforts. The transplant experiments with Trilliumgrandiflorum showed that, as with natural Trilliumpopulations, grazing impacts where deer occur at highdensity include a large reduction in flowering rates andsurvivorship within the growing season and a decline inplant size across years.

Stem damage by lepidopteran larvae also affectedtransplant survival, indicating that under reduced deerdensities other sources of damage could inhibit trans-plant success. These experiments show, however, thatat high deer density significant additive deer impacts oc-cur even when other important factors affect transplantsurvival and reproduction (H2, protected versus unpro-tected plants). Because our objective was to measuredeer impacts, we did not assess whether factors such asseed production and subsequent survivorship would besufficient for successful T. grandiflorum establishment.The observed levels of deer grazing, Trillium survivaland flowering rates, and changes in plant size at the low-deer-density site show that, given suitable conditions forT. grandiflorum establishment, deer at low density arenot expected to inhibit transplant success. Long-termmonitoring is needed to confirm this.

Our combined studies show that deer exert dramaticinfluences on Trillium reproduction and populationstructure at sites supporting high overwinter deer con-centrations (25–35 deer/km2). Levels of herbivory ob-served at two sites in southeastern Minnesota suggestthat deer can lead to the local extirpation of selectedforbs such as Trillium in forest fragments and can in-hibit efforts to restore populations of palatable species.The most significant effects on population structurewere observed where deer occur at high overwinterconcentrations for at least 5 years. Significantly lowergrazing intensities and higher Trillium flowering rateswere observed where deer occur at low overwinter den-

sity (4–11 deer/km2) due to annual hunting, but long-term monitoring is needed to determine whether thislevel of herbivory does not lead to Trillium declines. Itis important for managers to recognize that, due togrowing-season dispersal of deer into the surroundingagricultural landscape, the absolute number of deer re-sponsible for the documented grazing impacts may besubstantially less than the winter density, but that siteswith high overwinter concentrations still have compara-tively high summer deer density and grazing intensity.

Our results suggest that the relationship between deerand forest forb communities in southeastern Minnesotahas undergone a rapid change in response to landscapefragmentation and changing deer densities over the past150 years, such that selective grazing can lead to the lo-cal extirpation of sensitive species. As a result, active,long-term management to limit deer densities in parksand preserves through hunting or other effective meth-ods appears necessary for the conservation and restora-tion of fragmented forest communities in eastern NorthAmerica.

Acknowledgments

We thank P. A. Jordan for advice on all aspects of thestudy. We also thank T. Pharis for his help initiating thestudy and T. Pharis, T. Shay, and D. Tierno for assistancein the field. P. A. Jordan, P. Kotanen, and two anony-mous reviewers provided helpful comments on previousdrafts of the manuscript. Financial support was providedby the Minnesota Agricultural Experiment Station, theNational Science Foundation, and the Dayton-WilkieFund for Natural History. We also thank J. Moriarty andJ. Vorland for providing the aerial count data and J.Blackmer, A. Grannis, L. Gilette, and R. Lambert for per-mission to work at the various study sites.

Literature Cited

Agresti, A. 1996. An introduction to categorical data analysis. JohnWiley & Sons, New York.

Allison, T. D. 1990. The influence of deer browsing on the reproduc-tive biology of Canada yew I. Direct effects on pollen, ovules, andseed production. Oecologia 83:523–529.

Alverson, W. S., D. M Waller, and S. L. Solheim. 1988. Forests to deer:edge effects in northern Wisconsin. Conservation Biology 2:348–358.

Alverson, W. S., W. Kuhlmann, and D. M. Waller. 1994. Wild forests: con-servation biology and public policy. Island Press, Washington, D.C.

Anderson, R. C. 1994. Height of white-flowered trillium (Trilliumgrandiflorum) as an index of deer browsing intensity. EcologicalApplications 4:104–109.

Anderson, R. C., and A. J. Katz. 1993. Recovery of browse-sensitivespecies following release from white-tailed deer Odocoileus virgin-ianus Zimmerman browsing pressure. Biological Conservation 63:203–208.

Anderson, R. C., and O. Loucks. 1979. White-tailed deer (Odocoileusvirginianus) influence upon structure and composition of Tsugacanadensis forests. Journal of Applied Ecology 16:855–61.



Augustine, D. J. 1997. Grazing patterns and impacts of white-taileddeer in a fragmented forest ecosystem. M.S. thesis. University ofMinnesota, St. Paul.

Balgooyen, C. P., and D. M. Waller. 1995. The use of Clintonia borea-lis and other indicators to gauge impacts of white-tailed deer onplant communities in Northern Wisconsin, USA. Natural Areas Jour-nal 15:308–318.

Beals, E. W., G. Cottam, and R. J. Vogl. 1960. Influence of deer on veg-etation of the Apostle Islands, Wisconsin. Journal of Wildlife Man-agement 24:68–80.

Berner, A. H., and D. E. Simon. 1993. The white-tailed deer in Minne-sota’s farmland. Wildlife report 9. Minnesota Department of Natu-ral Resources, St. Paul.

Daubenmire, R. F. 1936. The “Big Woods” of Minnesota: its structureand relations to climate, fire and soils. Ecological Monographs 6:233–268.

Dexter, M. 1996. Status of wildlife populations, fall 1996 and 1982 -1995, hunting and trapping statistics. Report. Section of Wildlife,Minnesota Department of Natural Resources, St. Paul.

Eberhardt, L., and R. C. van Etten. 1956. Evaluation of the pellet groupcount as a deer census method. Journal of Wildlife Management20:70–74.

Frelich, L. E., and C. G. Lorimer. 1985. Current and predicted long-term effects of deer browsing in hemlock forests in Michigan, USA.Biological Conservation 34:99–120.

Gleason, H. A., and A. Cronquist. 1991. Manual of vascular plants ofnortheastern United States and adjacent Canada. New York Botani-cal Garden, New York.

Graham, S. A. 1952. Results of deer exclosure experiments in the Ot-tawa National Forest. Transactions of the North American WildlifeConference 23:478–490.

Grimm, E. 1984. Fire and other factors controlling the Big Woods vege-tation of Minnesota in the mid-nineteenth century. EcologicalMonographs 54:291–311.

Hanzawa, F. M., and S. Kalisz. 1993. The relationship between age,size, and reproduction in Trillium grandiflorum (Liliaceae). Amer-ican Journal of Botany 80:405–410.

Harper, J. L. 1969. The role of predation in vegetational diversity. InDiversity and stability in ecological systems. Brookhaven Sympo-sium in Biology 22:48–62.

Howard, W. E. 1966. Control of introduced mammals in New Zealand.Information series 45. New Zealand Department of Science and In-dustrial Research.

Jakes, P. J. 1980. The fourth Minnesota forest inventory: area. Re-

source bulletin 54. U. S. Forest Service, North Central Forest Exper-iment Station, St. Paul, Minnesota.

Jordan, P. A., R. O. Peterson, J. P. Campbell, and B. McLaren. 1993.Comparison of pellet counts and aerial counts for estimating den-sity of moose at Isle Royale: a progress report. Alces 29:267–278.

Kawano, S., M. Ohara, and F. H. Utech. 1986. Life history studies onthe genus Trillium (Liliaceae) II. Reproductive biology and survi-vorship of four eastern North American species. Plant Species Biol-ogy 1:47–58.

Leuthold, W. 1996. Recovery of woody vegetation in Tsavo NationalPark, Kenya 1970 - 1994. African Journal of Ecology 34:101–112.

Lubbers, A. E., and M. J. Lechowicz. 1989. Effects of leaf removal on re-production vs. belowground storage in Trillium grandiflorum.Ecology 70:85–96.

McShea, W., and J. Rappole, editors. 1997. The science of overabun-dance: deer ecology and management. Smithsonian InstitutionPress, Washington, D.C.

Miller, S. G., S. P. Bratton, and J. Hadidian. 1992. Impacts of white-tailed deer on endangered and threatened vascular plants. NaturalAreas Journal 12:67–74.

Ohara, M., and F. H. Utech. 1988. Life history studies on the genus Tril-lium (Liliaceae) V. Reproductive biology and survivorship of threedeclinate flowered species. Plant Species Biology 3:35–41.

Ownbey, G. B., and T. M. Morley. 1991. Vascular plants of Minnesota.University of Minnesota Press, Minneapolis.

Rogers, R. S. 1981. Mature mesophytic hardwood forest: communitytransitions, by layer, from East-Central Minnesota to southeasternMichigan. Ecology 62:1634–1647.

Rooney, T. P. 1997. Escaping herbivory: refuge effects on the morphol-ogy and shoot demography of the clonal forest herb, Maianthemumcanadense. Journal of the Torrey Botanical Society 124:280–285.

Spatz, G., and D. Mueller-Dombois. 1973. The influence of goats onkoa tree reproduction in Hawaii Volcanos National Park. Ecology54:870–876.

Strole, T. A., and R. C. Anderson. 1992. White-tailed deer browsing:species preferences and implications for central Illinois forests.Natural Areas Journal 12:139–144.

Tilghman, N.G. 1989. Impacts of white-tailed deer on forest regenera-tion in northwestern Pennsylvania. Journal of Wildlife Management53:524–532.

White, G. C. 1992. Do pellet counts index white-tailed deer numbers andpopulation change? Journal of Wildlife Management 56:611–612.

Vasilevsky, A., and R. L. Hackett. 1980. Timber resource of Minnesota’scentral hardwood unit, 1977. Resource bulletin NC-46. U.S. ForestService, North Central Forest Experiment Station, St. Paul, Minnesota.

Effects of White-Tailed Deer on Populations of an ......Effects of White-Tailed Deer on Populations...

Documents

Transcript of Effects of White-Tailed Deer on Populations of an ......Effects of White-Tailed Deer on Populations...